Help with biostatistical analysis for a study

[GreyGhostRos]

Hello.
I need some guidance with respect to biostatistical analysis of my study data:
Prospective study which studies the outcome of a surgical procedure done for a fracture with a sample size of 30 patients
The outcome is being measured as two categorical variables (functional and radiological) at the end of 6 months from surgery.
Both outcomes are measured as 4 categories (Excellent, Good, Acceptable, Poor).

I wish to analyse the data as:
1. Are the surgical results significant in terms of functional outcome and radiological outcome separately
2. Is there a correlation between the functional and radiological outcome

Which test should be applied as there is no population data?

Any help will be greatly appreciated
Thank You.

 

[GretaGarbo]

Waoh! Someone is asking for advice before the study. Applaud!
(Otherwise it is common: "I have some data.What am I going to do with it?")

I guess this surgical procedure is a new method. Do you have a control group with "traditional" surgical? (Otherwise the investigation is not so meaningful.) Can you randomize patients to treatment or control?

Do you have an idea about what would be a clinically relevant difference?

Have you thought about what kind of statistical model you would like to use?

 

[GreyGhostRos]

Yes. Traditional surgical method is available. But we are not planning to randomize cos the sample size at our center for such particular fracture is not very large. Think of it as a pilot study to assess the outcome (Standard questionnaire based scoring systems which assess the pain and function of the joint involved)
Problem with a comparative study with traditional methods is that multiple old studies have shown a varying range of quality of outcomes (surgical skill etc come into play).. So we don't have an ideal paper to compare our results against..

So what we wish to do is simply assess how many of our 30 patients have an excellent outcome, how many have a good outcome, and so on.. And this is where I am stuck, if there is a way we can assess the significance of our findings?

 

[GretaGarbo]

Well, if you have nothing to compare your results to, then the investigation is essentially meaningless. I hope that you prefer to hear such a frank statement now before the study, than to hear it afterwards.

If you have no idea about what a control group would have responded, what can you do with the numbers? Nothing, I would say.

It could be that your surgeons are more skillful, that your patients are in better shape from start or that they respond more generously.

So we don't have an ideal paper to compare our results against..

Usually one investigates the statistical significance by testing for example if the proportion "excellent" in the treatment group is statistically larger than in the control group. But if you have no control group, such a comparison is not possible.

 

[hlsmith]

Greta is right about the importance of randomization and comparator group!

Any loss to follow-up possible?

 

[GreyGhostRos]

@GretaGarbo,
Yes I understand that. That is exactly why I decided to ask here.. I have a good idea how to go about comparative type of studies but I could not get my head around what to do with such a situation.. I myself am wondering if there is anything called biostatistical "significance" out of this.. SO if i go with this design all I can speak of is percentage of excellent outcomes and such, right?
Yes.. I want a frank opinion so I asked. Better change the study structure before than after right?

@hlsmith,
Death pre or post operatively due to any cause will basically count as a loss to follow-up here..