Prediction factor for Chronicity - Which Test should I use?

#1
Hey All,

After a few hours of researching and comparing I'm still stuck. I don't know which I test I need to use for my topic. Heres the situtation:

The goal of my research is to see if the values of a certain questionnaire can have an influence on whether or not someone will develop chronic low back pain.

The questionnaire has list of questions that you can answer between completely disagree (0 points) to completely agree (5 points). Giving a range of 0-80 total possible points. We measured the participants pain levels in the acute stage (less than 4 weeks). And then again at 6 weeks, 12 weeks, 6 months and 1 year.

Ofcourse there will be some participants that are pain free at the 6 weeks mark and some that will still have pain after 1 year. Can this questionnaire be one (of the many) factor that could have an influence on the pain levels.

Let me know if you need more information.

Thanks alot!
 

Karabiner

TS Contributor
#2
Can this questionnaire be one (of the many) factor that could have an influence on the pain levels.
Do you want to include those other factors also into your analysis?
Is the outcome here "pain free after 1 year" yes/no?
How large is your sample size, by the way, how many are lost to follow-up, and how large is the proportion of
pain free / not pain free patients?

With kind regards

Karabiner
 
#4
Do you want to include those other factors also into your analysis?
Is the outcome here "pain free after 1 year" yes/no?
How large is your sample size, by the way, how many are lost to follow-up, and how large is the proportion of
pain free / not pain free patients?

With kind regards

Karabiner
No, these other factors are not included.

Maybe I can make it easy first by stating that the outcome is "pain free after 1 year" yes/no. But it would also be interesting to see if theres a correlation between higher score on the questionnaire and still having pain later. So if you score a medium score, you might be pain free after 6 month already and if you scored a higher score that you might only be pain free after one year, for example.

Sample size is about 300. There are no pain free patients. We only took patients with acute low back pain.

Thanks!
 

Karabiner

TS Contributor
#5
To test whether there is an association between score and pain status is simple, just compare the initial score between
the pain-free group and the chronic pain group (t-test or U test). Study drop outs could produce a bias here, though.
You could also have a look at boxplots (distribution of baseline scores for the pain free group at e.g. 6 week vs. baseline
scores for the pain experiencing group at week 6), and at ROC curves if you want to know whether there's a value for the
score which can reasonably discriminate between the 2 groups.

You could additionally compute a Spearman rank correlation between the score and the first pain-free follow-up assessment
("1" after 6 weeks, "2" after 12 weeks etc.; and if the patient is not pain free after 1 year, this is counted as "6").

With kind regards

Karabiner
 
Last edited: