# Thread: biostatistics for ordinal variable

1. ## biostatistics for ordinal variable

The study is perspective.
the subject is : the effect of an intervention on antibiogram level

I will have a antibiogram value at (4: for clean- 3 for a few- 2 for moderate and 1 for many bacteria) for each patient( 5 patient in every group, 10 patient in control and intervention groip).
and i have collected antio biogram for 4 time, base, after 10 days, after 20 days, after 30 days.
______

Question

My question is: how can I statistically assess whether or not there is a significant difference in the (hypothesised) antibiogram from base to 30 day between the study group and the control group, and which type of analysis and test come into play?
thank you so musch

2. ## Re: biostatistics for ordinal variable

You want a nonlinear multilevel model. "Nonlinear" because your dependent variable is ordinal, not continuous and "multilevel" because you have repeated measures and, therefore, violate the assumption of independence. These are complex models. They are available in SAS (GLIMMIX or NLMIXED) and R (nlme) and possibly other packages.

3. ## Re: biostatistics for ordinal variable

but i change ordinal values to numerical as 1 2 3 4 for many , mod and etc. any one could hepl me?

4. ## Re: biostatistics for ordinal variable

At each time point you could determine whether antibiogram value
has improved compared to baseline, and then compare the new variable
("improvement yes/no") between groups usin Chi2 or Fisher's test.

With kind regard

K.

5. ## The Following User Says Thank You to Karabiner For This Useful Post:

parisaarzani (11-13-2014)

6. ## Re: biostatistics for ordinal variable

i do this , but i have a problem, my antibiogram level has 4 status, many- mod- a few- cleam!!!

7. ## Re: biostatistics for ordinal variable

Originally Posted by parisaarzani
i do this , but i have a problem, my antibiogram level has 4 status, many- mod- a few- cleam!!!
Where exactly is the problem, then? You can tell, for every
single case, whether there was a change from baseline or not.

If patients were randomized into the respective groups,
you could alternatively just perform an U-test at each time
point (analysing the original dependent variable).

With kind regards

K.

8. ## Re: biostatistics for ordinal variable

the problem is here: if someone is clean at base because of antibiotic usage then after cut of antibiotic and doing our intervention , again the antibiogram is clean . what should we name this event?

9. ## Re: biostatistics for ordinal variable

the antibiogram has 4 level and sometimes the level is constant but it means it has an improvement

10. ## Re: biostatistics for ordinal variable

Originally Posted by parisaarzani
the problem is here: if someone is clean at base because of antibiotic usage then after cut of antibiotic and doing our intervention , again the antibiogram is clean . what should we name this event?
Unchanged. Or perhaps you should have inclusion criteria
for study subjects, such as: only subjects are included who
were un-clean at baseline. But such considerations depend
on substantial knowledge in the field, which I don't have.
the antibiogram has 4 level and sometimes the level is constant but it means it has an improvement
Again, what you consider a success depends on your own knowledge
and research interests.

With kind regards

K.

11. ## Re: biostatistics for ordinal variable

No, i think you didnet understand my question. because at first all patients were using antibiotics so the antibiogram is clean.

12. ## Re: biostatistics for ordinal variable

I see. So, at baseline, all partcipants have value "4"?
You can compare groups at each time point using a
U-test, then.

With kind regards

K.

13. ## Re: biostatistics for ordinal variable

Originally Posted by Karabiner
I see. So, at baseline, all partcipants have value "4"?
You can compare groups at each time point using a
U-test, then.

With kind regards

K.
i found the answer myself, i should use freedman test for trend analysis, thanks for all members attention

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts