Please help! Am I using the right SPSS statistical test?

[WilliamB]

Hello, I am looking into a study comparing isometric-eccentric contractions vs eccentric contractions for a knee condition.
Please advise on a statistical test(s) you think should be used to compare these two IVs against the outcome of the DV?
I will be testing the DV at baseline, 3, 6 & 12 months.

I was thinking a Mixed ANOVA in SPSS?

Thanks in advance!

 

[Karabiner]

It looks like you have 1 independent variable "contraction type" with 2 levels. And, possibly, "time" in addition.

The statistical analysis depends on the nature of your dependent variable, your exact research question(s),
the sample size(s), and whether patients were lost between baseline and 12 months.

With kind regards

Karabiner

 

[WilliamB]

Thanks for your reply.
So I guess I am confused between my Independant Variables. To clarify, I will not be consucting this study. I am writing a grant proposal as if I were going to perform this study.

The dependant variable is an outcome measure, measuring pain and functional ability of the condition.

The research question is ‘Are isometric-eccentric contractions superior to eccentric contractions as treatment for patellar tendinopathy, a randomised control trial’ - may I ask why does the question matter?

I do not know the sample size yet because I cannot do a calculation of power for sample size using G*Power without knowing the statistical test.

Let’s suppose no patients would be lost.

I’d appreciate some feedback or a test recommendation.

 

[ondansetron]

The research question is ‘Are isometric-eccentric contractions superior to eccentric contractions as treatment for patellar tendinopathy, a randomised control trial’ - may I ask why does the question matter?

I don't think you've properly defined the question to a formalized hypothesis. How are you going to judge superiority?
The bold text is also a giant problem in medicine that comes up often-- we want to know how to analyze the data without knowing clearly the question being answered. The question is probably the most important factor (but I'll settle for second most) in determining how to analyze the data (try to answer the question); this is why @Karabiner asked. Asking "why does the research question matter" is the same as asking why the chief complaint and H&P matter for diagnosing and treating a patient who came into clinic.

 

[Karabiner]

The dependant variable is an outcome measure, measuring pain and functional ability of the condition.

How does it exactely look like? There exist different levels of measurement; categorical
like "dysfunctional versus functional"; ordered categorical like "no pain - mild pain - severe pain -
unbearable pain"; or interval scaled like " ° knee flexion". For different levels of the outcome
variable(s), different analyses are possible.

The research question is ‘Are isometric-eccentric contractions superior to eccentric contractions as treatment for patellar tendinopathy, a randomised control trial’ - may I ask why does the question matter?

How is superiority defined - which observations will show you whether one treatment is superior?
e.g. do you expect superiority already at 3 months or only in the long run?

By the way, is allocation to groups randomized, or are these natural groups?
Will additional patient variables be measured which could be used as co-variables
when predicting outcome and/or which could be used a strata (if patients are
randomized)?

Let’s suppose no patients would be lost.

Normally, a study application has to inform the reader how drop-outs will be dealt with.
It is not conceivable that all patients will complete a study over 12-months. The question
is, whether the rate of loss to follow-up is usually low in this kind of treatment study.
But anyway, if you assume that your reviewers will not mind, then we'll ignore this issue.

With kind regards

Karabiner

 

[WilliamB]

How does it exactly look like?

It is a VISA scale that measures pain and functional ability on a scale out of 100 - it is categorical/ordered-categorical. Please find the attachment below if you wish to see it.

How is superiority defined - which observations will show you whether one treatment is superior?

I do expect to see superiority after the intervention period at 3 months and continued to 12 months by the training group. Should I include when I assume superiority (3months) in the title then?

The participants are to be randomly assigned, through the RCT. I could additionally also measure their vertical jump height as a secondary outcome measure.

Going back to sample size, am I right in thinking I cannot predict the appropriate size without my statistical test?

Thanks a lot for your help, I am slightly confused still..!

 

[Karabiner]

It is a VISA scale that measures pain and functional ability on a scale out of 100 - it is categorical/ordered-categorical.

There's been some discussion whether it can be treated as interval scaled (which would make things much easier).
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3805376/

I do expect to see superiority after the intervention period at 3 months and continued to 12 months by the training group. Should I include when I assume superiority (3months) in the title then?

Commonly you have to define a clear primary end point,
i.e. which single measure (or composite of several measures)
at which point in time is used to demonstrate a difference.
https://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm536750.pdf

Other analyses may also be important and therefore be included
into the study protocol, as secondary end points. So, your primary
endpoint could be "pain measured at month 3", for example.

If the VAS can be treated as quasi-interval scaled, a mixed ANOVA
could be used. If the VAS is ordinal, then things will become
more complicated. Personally, I have used GEE (generalized estimating
equations) for this kind of design.

Going back to sample size, am I right in thinking I cannot predict the appropriate size without my statistical test?

Yes, you need to know the analytical strategy, and the particular test.

With kind regards

Karabiner

 

[WilliamB]

Commonly you have to define a clear primary end point,

So would this be a more effective title? - What effect do Isometric-Eccentric contractions have on VISA score and jump performance in Patellar Tendinopathy from a 12-week intervention? A Randomised Control Trial.

a mixed ANOVA

So are you suggesting i conduct the mixed anova? What about one way/within/between interactions?

Thanks