Evidence Based Practice: Research, uncertainty and
decision-making
I want to thank Erson Religioso III (@The_OMPT) for
bringing this
study to my attention to give an evidence based practice perspective
on it. It is a great study that took two rather large existing cohort data sets
(n=696 and n=336) to look at the correlation of pain with grades of
osteoarthritis, osteophytes and joint space narrowing in knees. Their
conclusion was that radiographic features of osteoarthritis were strongly
associated with knee pain. Obviously many would think this would be a dramatic
blow to a “pain science” person like myself. I can hear the naysayers: “Isn’t
this study PROVING that pain and injury are the same and all that talk about
the brain and pain is worthless?” “Doesn’t it PROVE that you have to just treat
the tissue dysfunction and pain will go away, because it is the tissue
dysfunction causing the pain?”
one study can show us something is not true (i.e. falsify), but it cannot definitively prove something is true
When critically looking at this study, or any study, we need
to keep a few general things in mind that are critical to any good analysis of
what research might be telling us and what it does not tell us. First, always
remember one of the hallmarks of correlational studies: Correlation does not
mean causation! Just because two things are correlated (pain and radiographic
findings), it does not mean that radiographic findings of OA cause pain. Second,
no one study can prove anything. Research is meant to falsify things not prove
things. One
study can show us something is not true (i.e. falsify), but it cannot
definitively prove something is true. We all love to pick a study “to prove”
we are right, can we say “hello confirmation bias”. We need to look at the body
of literature (multiple studies) to help us come to a more correct or probably
stated better a less wrong conclusion.
This tells me this topic (pain and knee arthritis correlation) is obviously not a black and white answer. Welcome to research and uncertainty
So how do I interpret this study? We know this study conflicts
with some other studies that have shown only moderate correlation with
radiographic findings of osteoarthritis and pain as they point out in their
introduction (see references 2-6). This tells me this topic (pain and knee
arthritis correlation) is obviously not a black and white answer. Welcome to
research and uncertainty, I will touch on this at the end of the post and what
that might mean for clinical decision-making. I think this study did find
similar results with other studies (this means it strengths the chance of the
conclusion being true) that mild levels of arthritis have a moderate correlation
with pain. As you look at the confidence interval of the lowest levels of the
Kellgren and Lawrence grade the range is on both sides of 1.0 so there is no correlation.
However, the higher the KL grade the increase in the correlation of knee
arthritis findings and pain. Notice though even with the higher correlation
scores at KL grade 4 there was still a large range in the 95% confidence
interval (this is the range number in the parentheses) which tells me there is
a low level of precision with the odds ratio. Large ranges are common with
small data sets – this data set had over 1,000 subjects, not so small, so again
more uncertainty. This is a case that while findings are significant according
to the p-value, the effect size tells another more important story we need to
pay attention to.
Remember pain neuroscience tells us tissue information is some of the input in to the brain, but tissue information is never a sole pain generator.
So what does this mean for me clinically? This gives me
information to share with a patient that should help them decide on treatment. If
the person has low levels of arthritis, I can tell them there is no correlation
with the radiographic findings and pain. So if they are worried about the pain,
the radiographic information is not very useful to use in deciding treatment
direction. The more the radiographic findings show more significant osteoarthritis
the more likely the arthritis we see is associated with the pain they are
experiencing, but even then, it is not a pure black and white answer. We could
say it is a bit more complex, which is usually the case when it comes to pain. Some
patients scored at the highest level of osteoarthritis and had no pain. Remember
pain neuroscience tells us tissue information is some of the input in to the
brain, but tissue information is never a sole pain
generator. All of this is important information to share with a patient to
assist in decision-making on treatment direction.
To finish lets go back to research and uncertainty. I think
many look to research to add certainty to their decisions they make in the
clinic. While research should direct us to make better decisions, decisions
that are less wrong. We live in a very complex and inter-related world and
research cannot answer all the questions that are out there, yes uncertainty.
So how does a clinician make decisions for their patient, if they cannot be
certain? Maybe we should not be making the decision; maybe it should be the
patient making the decision after we fully inform them of everything?
Okay your turn: What is your take on the study and who
should make the decision on treatment - the clinician or the patient? – What say you?
image credit
Special thanks to the newest member of The Manual Therapist blogging team, Dr. Kory Zimney, PT, DPT, CSMT, CAFS
Keeping it Eclectic...
Special thanks to the newest member of The Manual Therapist blogging team, Dr. Kory Zimney, PT, DPT, CSMT, CAFS
Interested in live cases where I apply this approach and integrate it with pain science, manual therapy, repeated motions, IASTM, with emphasis on patient education? Check out Modern Manual Therapy!
Keeping it Eclectic...
 for bringing t...){kind=link}
Post a Comment
Post a Comment