Help needed in Clinical trial data with Minitab

[JUSTICE]

Hi to all.

We are testing a pain relief drugs. There are 3 drugs, T1,T2 and T3. The treatment T1 is a form of control and is a standard drug. there are 43 patients in the study. Each patient can only receive any 2 of the treatments one at period 1 and another at period 2.
The patients in the first two period are different from the patients in the second two periods. Example, the numbers of pain-free hours for the first patient at period 1 and 2 is 2 and 10 respectively and for patient 7 is 4 and 4 respective.
The responses are the numbers of pain-free hours after taking any of the drugs.

Patients

Obs Period Drug 1 2 3 4 5 6 7 8

1 1 T1 2 6 4 13 5 8 4 .
2 2 T2 10 8 4 0 5 12 4 .
3 1 T2 2 0 3 3 0 . . .
4 2 T1 8 8 14 11 6 . . .
5 1 T1 6 7 6 8 12 4 4 .
6 2 T3 6 3 0 11 13 13 14 .
7 1 T3 6 4 4 0 1 8 2 8
8 2 T1 14 4 13 9 6 12 6 12
9 1 T3 12 1 5 2 1 4 6 5
10 2 T2 11 7 12 3 7 5 6 3
11 1 T2 0 8 1 4 2 2 1 3
12 2 T3 8 7 10 3 12 0 12 5

I tried using Minitab to do the analysis but it seems am using it wrongly (perhaps the way I captured the data) because I am getting the total degree of freedom to be 85 and I thought it should b 42.
This is how I captured the data into Minitab:
Period Treatment Number of pain-free hours
1 T1 2
2 T2 10
1 T1 2
2 T2 8
1 . .
2 .
1 . .
.
.
.
1 T2 3
2 T3 5

I used the time period as blocks, so I have 1 degrees of freedom for the block and 2 degrees of freedom for the treatment and 85 for the total which I think is wrong.

Please, I need a help .

I will appreciate any support given me.
Thank you.

 

[trinker]

I suggest you check out this page(it's for R not Minitab but the concept is what we're looking at) on repeated measures (which I assume is what's going on in your study) that you can see from the output what the df should be. I think this will clear it up for you. Let us know either way?


(LINK)

 

[JUSTICE]

Ok. I will and will get back for any further assistance.
Thanks

 

[JUSTICE]

I read through many other documents in the link but it seems my own is a special case. All the repeated measures I have seen in the link you gave me were carried out in a way that each subject received same treatment for the same number of times. However, in my studies, subjects received same number of treatments (2 treatments each) but some have different treatments (eg. first patient had treatment T1 AND T2 while the 43rd patient had treatment T2 AND T3).

Please, more help needed here.
Thanks

 

[Dragan]

I read through many other documents in the link but it seems my own is a special case. All the repeated measures I have seen in the link you gave me were carried out in a way that each subject received same treatment for the same number of times. However, in my studies, subjects received same number of treatments (2 treatments each) but some have different treatments (eg. first patient had treatment T1 AND T2 while the 43rd patient had treatment T2 AND T3).

Please, more help needed here.
Thanks

My first question would be is: Why did you design your study in the manner that you did?...From an experimental design point of view-- your design appears to be poorly constructed.

 

[JUSTICE]

Dear Dragon,
The design was done to compare the 2 new drugs (T2 and T3) with a standard drug (T1) which is a form of control. A patient can only take any 2 of the treatments with one at each time period. The drug in question is a sleeping drug. So after each treatment, the number of sleeping-hours is recorded. Could this be a poor design? Our interest is to compare the means of the 3 treatment for their significance.
I believe there could be a way to handle this problem. We assume the interval of the time period is good enough to avoid carry-over effect.

Please, somebody should come to my aid.
Thanks.