Hi All
As a user of ultrasound and having just done a critical analysis on its
effects for part of my Masters degree I thought I would point out some
information I think is relevant.

Research Problems
Much of the research on ultrasound that has been reviewed during the process
of systematic reviews and meta-analyses in the past, is reported to be of
low quality (Gam & Johannsen, 1995; van der Windt et. al., 1999). This poor
level of quality could be due to the many methodological problems and
differences seen in the reviewed studies. To improve the quality of future
studies, and therefore the validity of the results, these problems and
differences need to be addressed.

Population Homogeneity
One of the largest problems with previous ultrasound research is the lack of
population homogeneity. Many of the studies analyse and compare the
population demographics, but most fail to categorise populations according
to a diagnosis, using broad descriptions instead.  This means that prior to
the delivery of the treatment intervention the study groups are not
identical, resulting in confounding errors (Vickers, 1995). Diagnosis has
been defined as a "statement that specifies the anatomical location and
specific pathologic classification of a disease process" (Laslett, 1996).
The use of terms such as ankle distortions and shoulder pain, as found in
studies assessed by Gam & Johannsen (1995), fail to meet the second aspect
of the definition.

If the disease process is not homogenous it would suggest that different
methodologies will be needed to treat the individual subjects in the study.
When individual studies fail to have homogenous study populations then
meta-analyses and critical reviews using terms such as 'musculoskeletal
conditions' will have even greater population heterogeneity and lead to
findings and conclusions that have little validity due to large confounding
errors.

Some systematic reviews subcategorise and group the studies into like
conditions such as 'shoulder pain' (van der Windt, 1999). Included under
this term were studies on subacromial bursitis, shoulder pain, shoulder cuff
lesions, and painful shoulders with hemiplegia, which are far from
homogenous pathologies. Subgroup analysis can have an important part to
play, but the classification has to be appropriate. Disease classifications
for a region would be appropriate, as they take the pathology and tissue
type into account. This improves group homogeneity and, therefore, improves
the validity of the results. A study of head and neck pain demonstrated this
after concluding that conditions with different underlying causes of pain
responded differently to treatment and, therefore, had distinct treatment
outcomes. (Lundeberg, et. al., 1988).

In physiotherapy treatment it is the disease process that is treated.
Grouping subjects in an experiment by the disease process, despite having
different anatomical locations, would be a better method to improve group
homogeneity. Johanssen, Gam & Karlsmark (1998) have shown that it is
possible to group by disease process, as compared to anatomical location.
They conducted a meta-analysis on the effect of ultrasound treatment on
healing, in which they demonstrated a statistically significant improvement
in healing for those who received treatment.


Generalising Conclusions
Many claims are made with respect to the efficacy of ultrasound. Article
titles claiming that 'ultrasound has no anti-inflammatory effect' (Goddard
et. al., 1983) need to be put into context. In the current literature,
studies on ultrasound have many methodological differences. These include
different frequencies, pulse ratios and intensities available. The delivery
is also dependant on other variables such as what type of coupling medium is
used, whether the ultrasound head is moved or stationary, and whether the
machine output is spatial average / temporal average (SATA) or spatial
average / temporal peak (SATP). Generalised conclusions made about the
efficacy of a modality when there are different methodologies used in one
study, or when the results of only one methodology are extrapolated to all
methodologies, should be performed with extreme care. Conclusions would be
more valid if they were limited to the population examined and methodology
used.


Study Controls - Ultrasound
Ultrasound studies should be relatively easy to control, as when delivering
a low dose no sensation is experienced. In circumstances when higher doses
are used and heating is experienced, the subject could be warned that there
may be some warmth. The application of a placebo treatment, therefore, only
requires some sort of disruption of the supply to the treatment head of the
machine. Any bias by the therapist can also be controlled by blinding or
using a technician, as once the dose is determined the delivery requires no
technical skill. Despite this many studies on ultrasound do not use placebos
as controls, and instead compare ultrasound to other treatments. In doing
this, the conclusions only establish if one treatment is better than the
other, and not if either is an effective treatment. The exception to this is
if the compared treatment has already been shown to be effective. Reasons
for comparing treatments may include the ethical dilemmas of withholding
treatment to a patient by the delivery of a placebo (Vickers, 1995), or
possibly that the research has been conducted by clinicians wanting to know
what treatment works best. If comparisons are made to other standardised
treatments such as a drugs, then it is important that the outcome measures
for the treatments are appropriate as their method of action may different
(Dyson, 1987). The use of animal studies, and experimental models as used by
Snow & Johnson (1988), go some way in establishing the effect of ultrasound
without these ethical problems of withholding treatment. The relevance,
however, of these type of studies to clinical practice is yet to be decided.

Conclusions on Research Problems
Current ultrasound research is in need of good quality controlled trials.
Until there is an improvement in quality and methodological problems are
addressed, the conclusions of systematic reviews and meta-analyses should be
interpreted with care. In particular where the study groups are not
homogenous, the placebo or control is not relevant for the situation, and
treatment delivery is not consistent throughout the study populations.
Without these factors being incorporated into study designs, any conclusions
on the efficacy of ultrasound will be limited.

Basically What I am saying is that if US was a drug then there would be
major problems getting ethical approval for these trials as there is such a
wide variety of dosages and conditions used. Imagine getting a heart drug
from you Dr and him guessing at the dose.!!!!
I have also had a quick look at the PT journal articles and not that

Johannsen, F., Gam, AN., & Karlsmark, T. (1998). Ultrasound therapy in
chronic leg ulceration: a meta-analysis. Wound Repair & Regeneration, 6(2),
121 - 126.

which is a meta-analysis with a positive result does not appear.

I therefore conclude that instead of "there is evidence that US has no
clinical effect" I say that "there is little or no evidence available that
US has a clinical effect" ie the evidence is not available yet

Hamish Ashton MHSc. Physiotherapy (A post grad degree with lots of clinical
involved for the DPT discussion)
New Zealand        Full references available on request