Welcome to the Podiatry Arena forums, for communication between foot health professionals about podiatry and related topics.
You are currently viewing our podiatry forum as a guest which gives you limited access to view all podiatry discussions and access our other features. By joining our free global community of Podiatrists and other interested foot health care professionals you will have access to post podiatry topics (answer and ask questions), communicate privately with other members (PM), upload content, view attachments, receive a weekly email update of new discussions, earn CPD points and access many other special features. Registered users do not get displayed the advertisments in posted messages. Registration is fast, simple and absolutely free so please, join our global Podiatry community today!
If you have any problems with the registration process or your account login, please contact contact us.
I have recently reveiwed several manuscripts that I recommended that editors not publish due to a fundamental flaw in the methodology. It concerned me enough to post a thread here about it (and will freely admit that I have been guility of this in the past, but times change as we learn more).
One potentially appealing thing about doing foot or podiatry research is that each subject has two feet, meaning that if you use both feet in the data, you have either doubled your sample size or halved the number of subjects used.
HOWEVER, a key assumption of almost all statistical tests is that the subjects in the sample are independent of each other ..... this means that you can not use two feet from the same person in the sample as they are related (not independent of each other; they are paired) - they have the same body weight; the same blood supply; etc etc ...
The use of the two feet of one subject is no longer acceptable in research due to this lack of independance. This is a common issue in the opthalmologic literature (two eyes or one eye?); the orthopaedic literature (two limbs or one?); the rheumatological literature (eg one knee or two):
Why are researchers still using both feet; still submitting the data for publication using both feet in the analysis; and why are journal editors still permitting them to be published (esp in podiatric journals)?
Have seen this problem myself in a number of manuscripts that I have reviewed recently too. What we used to do was to test left foot data against right foot data for statistical significance and then pool the data if it was not different.
Playing Devil's advocate: As by their very nature left and right feet are opposite, are they really paired data? The argument that they have the same body weight, the same blood supply etc, doesn't really hold true either. How dissimilar does something have to be before it can be considered not paired? Taking this argument to completion, if the researchers first demonstrated statistical difference between the left and right data could they then pool the data? This would mean that we have a binomial population and the assumptions of normal distribution would be violated, but the good old t-test is pretty robust isn't it.
Thanks for the above, this one was bugging me for quite some time and you could see that there was no consensus on the topic. I can imagine that it's going to take a while before the pooling of data duel is going to end...
What the most important aspect of this discussion is (should be), is that podiatric research should take a stand and stick with it.
__________________
kind regards
Ken
Ken Van Alsenoy, MSc Pod
Artevelde University College
Podiatry dept.
Ghent - Belgium My location
Being that in the human body, both feet are not identical, why not consider each foot as a subject? true they are not 'strictly' independant, being that they are fromteh same person and therefore similar, but then again mosat of the 'RCT's do not really have a randomised population, it's usually a convenience sample, or a clinically controlled trial, or whoeve happens to walk in. Are we clutching at straws??
__________________
Adrian Misseri
B.Pod.,M.Hlth.Sci.(Pod.)
Playing Devil's advocate: As by their very nature left and right feet are opposite, are they really paired data?
Quote:
Originally Posted by Kenva
What the most important aspect of this discussion is (should be), is that podiatric research should take a stand and stick with it.
Quote:
Originally Posted by Adrian Misseri
Being that in the human body, both feet are not identical, why not consider each foot as a subject?
Other disciplines have made that decision. You would not get away with using two eyes in the opthalmological literature; you won't get away with using two knees in the rheumatological literature; you won't get away with it in the biomechanics literature; why is that you can still get away with it in the podiatric literature? Why should podiatric research be at a 'lower' level than other disciplines?
So in doing research, do we need to specify if it's left or right foot? Or can we just ramdomly pick a foot?
Fom all my understanding, it does not matter. You can use just the left or right foot; randomly select one foot; use the symptomatic foot; use the most symptomatic foot.
A while ago one of my Hons students did an intensive search to see what was more appropriate - ie should he use the left in all subjects or the right in all subjects; or randomly select the left or right. He could not find any convincing arguments for either approach (which is what I told him in the first place )
Incidently, there are situations in which data from both feet is needed (eg if you are doing a study comnparing the left side to the right side, but the statistical prerequisites are not breached by that kind of analysis)
So in doing research, do we need to specify if it's left or right foot? Or can we just ramdomly pick a foot?
First I would want the researcher to demonstrate that there is no statistical difference between the left and right foot data for the variables under investigation. If no differences were identified, then it would be ok to randomly select right or left by the flip of a coin. This could be for the population or subject by subject. But since they would need to show no difference in the first place they might as well go on to examine both right and left foot data independently! And if they do find differences.... is it really paired data?????? See my previous post.
Think about this: lets say we measured HA angle in twenty subjects (40 feet) and the mean HA angle in the left foot was 5 degrees (3.5, 6.5) and in the right foot it was 15 degrees (12, 17) would it be safe to randomly select one foot for analysis over the other? If I choose to analyze the right foot data only what bias would this introduce? Adrian's point re: selection is well made
Craig has asked me to add my two cents worth to the discussion on this issue, which was initially brought to my attention by Lloyd Reed at QUT. My view on this is explained in two papers:
Menz HB. Two feet, or one person? Problems associated with statistical analysis of paired data in foot and ankle medicine. Foot 2004;14:2-5.
Menz HB. Analysis of paired data in physical therapy research: time to stop double-dipping? J Orthop Sports Phys Ther 2005;35:477-478.
There's fair degree of overlap between these two papers, as the second one was an invited editorial in which I was specifically requested to briefly summarise the key points of The Foot paper.
I've uploaded the PDF of The Foot paper (and a reply by Dr Tony Redmond) for those interested in exploring this issue in more detail.
As a general rule, pooling data should be avoided for the reasons outlined in the paper. However, in some cases it may be justified, particularly in situations where the variables being examined are very foot-specific, there is a high degree of variation between feet, and the condition being examined is frequently unilateral.
I guess the bottom line is that authors who decide to pool data from both feet should provide a conceptual and statistical justification for doing so, rather than just double-dipping to increase sample size (which I suspect is a very common reason for doing it).
The Following 2 Users Say Thank You to Hylton Menz For This Useful Post:
Fom all my understanding, it does not matter. You can use just the left or right foot; randomly select one foot; use the symptomatic foot; use the most symptomatic foot.
Just at ESM mtg in Dundee and in one of the papers from todays session the reseachers used another option: they used the dominant limb (by getting the participant to nominate which limb they preferred to stand on in a one legged stance position). I guess this option is fine unless the research question's answer is affected by limb dominance issue.
Statistical Consideration for Bilateral Cases in Orthopaedic Research
Moon Seok Park, Sung Ju Kim, Chin Youb Chung, In Ho Choi, Sang Hyeong Lee, and Kyoung Min Lee J Bone Joint Surg Am. 2010;92:1732-1737
Quote:
Background Statistical independence means that one observation is not affected by another; however, the principle of statistical independence is violated if left and right-side measures within a subject are considered to be independent, because they are usually correlated and can affect each other. The purpose of the present study was to analyze the violation of statistical independence in recent orthopaedic research papers and to demonstrate the effect of statistical analysis that considered the data dependency within a subject.
Methods First, all original articles that had been published in The Journal of Bone and Joint Surgery (American Volume) over a two-year period were evaluated. The analysis was designed to identify articles that included bilateral cases and possible violations of statistical independence. Second, a demonstrative logistic regression without consideration of statistical independence was performed and was compared with a statistical analysis that considered data dependency within a subject. Radiographs of 1200 hips in 600 patients were used to examine the differences in terms of odds ratios (with 95% confidence intervals) of the risk factors for hip osteoarthritis.
Results Four hundred and eighty-six original articles were reviewed, and 151 articles (including forty-one articles involving the hip, thirty-four involving the knee, twenty-one involving the foot or ankle, nineteen involving the shoulder, ten involving the hand or wrist, nine involving the elbow, and seventeen involving other structures) were considered to include bilateral cases. Of the 486 articles that were reviewed, 120 articles (25%) (including thirty-six articles involving the hip, twenty-six involving the knee, fifteen involving the foot or ankle, fourteen involving the shoulder, seven involving the elbow, six involving the hand or wrist, and sixteen involving other structures) were found to have possibly violated statistical independence. Demonstrative statistical analysis showed that logistic regression was not robust to the violation of statistical independence. The 95% confidence intervals of the odds ratios for the risk factors showed narrower ranges (1.13 to 2.68 times) when data dependency within a subject was not considered.
Conclusions Researchers need to consider statistical independence when performing statistical analysis, particularly in studies involving bilateral cases. If data dependency within a subject is not considered, studies involving bilateral cases can bias results, depending on the context of those studies.
If I may take this space to voice my own personal vendetta among the majority of orthoses research:
Far too many studies are donw with the subjects wearing their own, usually-way-too-worn shoes. How in the world would anyone achieve reasonable levels of control under this kind of situation?
__________________ Jeremy Long C Ped
Smoky Mountain Foot Clinic
Far too many studies are donw with the subjects wearing their own, usually-way-too-worn shoes. How in the world would anyone achieve reasonable levels of control under this kind of situation?
Its actually quite easy to control for this.
The whole point of using randomization is that you end up with 2 groups that are theoretically identical in all their characteristics (ie age, weight, foot type, etc). Each group theoretically will have equal proportion of good and bad shoes in them. Many RCT's to stats tests on a whole range of baseline characteristics to show the groups are teh same.
As you have two "identical" groups, you then apply an intervention to one group and follow them over time - this means the groups are the same except for one characteristic (the intervention) ---- you can then determine what the effects of that intervention was.
If you also intervened to change the footwear as well, then what actually worked? Was it the footwear change or was it the orthotic change?
UNLESS, your research question was actually about what are the combined effects of footwear and orthotic change, in which case, you do change the footwear. If your research question was just looking at orthotic's then you do not change the shoes.
Pragmatically, we have to work clinically with people who do not change their shoes, so the RCT's are just reflecting that.
Research on one foot, two feet, or one person
)
Linear Mode
